Ramon Loya

Native English speaker with hands-on experience red-teaming conversational AI systems through a production-grade adversarial testing platform (RTK-1). Experienced in executing jailbreaks, prompt injections, misuse cases, bias exploitation, and multi-turn adversarial strategies. Skilled at annotating failures, classifying vulnerabilities, flagging systemic risks, and documenting findings reproducibly using structured taxonomies, benchmarks, and testing playbooks.

Strong analytical and adversarial thinking skills with proven ability to work independently in a remote, project-based environment.

Red Teaming Professional at AI Red Teaming Toolkit (RTK-1) – Personal Production Platform (2025-01 – Present)

Red-teaming conversational AI systems using advanced adversarial techniques and structured frameworks.

• Red-teamed conversational AI systems using jailbreaks, prompt injections, misuse cases, bias exploitation, and Crescendo multi-turn adversarial strategies with PyRITProvider and LangGraph orchestration
• Generated high-quality data by annotating failures, classifying vulnerabilities, and flagging systemic risks using DeterministicScorer and MultiJudgeConsensus (3-judge LLM voting with HITL on low confidence)
• Followed structured taxonomies, benchmarks, and testing playbooks covering OWASP LLM Top 10 and MITRE ATLAS frameworks
• Documented findings reproducibly through reports, datasets, and adversarial test cases including compliance-mapped PDF reports, JSON audit logs, and CampaignHistory with ASR metrics
• Evaluated AI outputs across sensitive or high-risk scenarios using SemanticDriftMonitor and BehavioralFingerprint for bias, misinformation, and harmful behaviors
• Demonstrated excellent written communication and documentation skills while working independently in a remote, project-based environment (FastAPI router, immutable domain models, autonomous CampaignScheduler, Prometheus metrics)

Systems Engineer at Homelab Infrastructure & Systems Troubleshooting Projects (2024-01 – Present)

Applied analytical and adversarial thinking skills to identify and resolve system weaknesses in infrastructure and networking.

• Applied strong analytical and adversarial thinking skills to identify and resolve system weaknesses in Proxmox VE cluster installations, Windows 11 clean deployments, and network/DNS configurations
• Performed detailed root-cause analysis (Safe Mode vs. clean boot isolation, log examination with dmesg/rsyslog, BIOS/UEFI/Secure Boot/TPM adjustments) while documenting every step for reproducibility and auditability
• Demonstrated ability to apply structured evaluation frameworks by systematically testing variables, validating changes incrementally, and producing clear technical documentation equivalent to incident reporting

Bachelor of Science in Cybersecurity and Information Assurance – Western Governors University (2024-01 – 2028-01)